Influence of fatness, intelligence, education and sociodemographic ...

3 downloads 0 Views 836KB Size Report
and increasing intelligence test score, educational level, current social class, age (up to 50 years) and proximity of ... having a higher frequency ofchronic diseases and by originating .... 1977-82 was obtained from National Health Service. Hospital ..... 3 Horwitz 0, Wilbeck E. Effect of tuberculosis infection on mortality risk.
Journal of Epidemiology and Community Health, 1989, 43, 369-374

Influence of fatness, intelligence, education and sociodemographic factors on response rate in a health survey STIG SONNE-HOLM, THORKILD I A SORENSEN, GORM JENSEN, AND PETER SCHNOHR From the Obesity Research Group, Hvidovre University Hospital, Copenhagen; and The Copenhagen City Heart Study, Rigshospitalet, Copenhagen, Denmark. ABSTRACT

Study objective: The aim was to investigate the characteristics of non-responders to an invitation to attend a health examination. Design: Taking advantage of an ongoing study of obesity, this was a survey of a cohort of severely obese men, with a randomly selected control group. Participants: The participants were draftees to the compulsory Danish military draft board examination between 1943 and 1977. Among 362 200 draftees, 1940 were identified as severely obese (body mass index > 31 kg/m2). A comparison group of 1801 subjects was randomly drawn from the remaining population. During the period 1981-3 those still alive and living in the same region (1651 obese, 1504 control) were invited to a health examination. Measurements and main results: The examination was attended by 964 obese (58%) and I 134 controls (75%). In both groups an increasing response rate was associated with decreasing body mass index, and increasing intelligence test score, educational level, current social class, age (up to 50 years) and proximity of residence. Logistic regression analysis showed that all these variables had independent effects on response rate. Frequency and duration of hospital admissions during the period 1977-82 did not differ among responders and non-responders in either group. Conclusion: Response rates in health surveys are strongly influenced by degree of fatness, intelligence, educational level, social class, age, and proximity of residence. Assessment of those who do not participate in health remaining population. The response rate was analysed surveys is important for two reasons: it may reveal bias in relation to information systematically gathered on in the results of the survey, and it contributes to the body mass index, intelligence test score and level of evaluation of the effectiveness of the programme in education at the time of appearance before the military reaching the target population. In longitudinal follow board, as well as on age, social status, residence and up studies, non-responders are characterised by previous hospital admissions at the time of follow up. having a higher frequency of chronic diseases and by originating from lower social classes than Methods responders. 1-3 Information on the non-responders can usually be obtained only from public registers unless All Danish men attaining the age of 18 years are the subjects are visited in their homes.45 In an registered with the military authorities and are epidemiological study on obesity, we had the examined by the medical board within the next few opportunity to explore in more detail the psychosocial years. The number and records of volunteers are not characteristics of non-responders. The study was available, but, according to the authorities, they based on a population of conscripts who had comprised about 2% during the period of this study. The study population comprised the 400 975 men undergone an obligatory examination several years prior to the present examination, to which we invited processed by the board in the metropolitan area of the most obese and a randomly selected group of the Copenhagen and surrounding counties from 1943 to 369

Stig Sonne-Holm, Thorkild I A Sorensen, Gorm Jensen, and Peter Schnohr 370 1977 and in the remaining part of Sjlland and modification of the scale, which was based on prestige surrounding islands from 1964 to 1977. All underwent in the society.'2 13 systematic examination, including measurement of height and weight, except for 4-5% who were FOLLOW UP STUDY unquestionably unfit for service and did not appear All subjects included in the obese group and a random before the board. Records were not available for those 50% sample of the control group, totalling 3707, were examined in the metropolitan area and found fit for selected for participation in the follow up study. service from 1958 to 1968 and who were living in During the period November 1, 1981 to August 15, another region in 1969. Thus, 362 200 were available 1983 information was obtained every third month for the study. The population has been described in from the Central National Register regarding current address of those for whom the examination was detail elsewhere.68 planned to take place during the ensuing three months. Only subjects living east of Store Belt at the time of OBESE GROUP Obesity isdefined here asabodymassindex > 31 kg/M2, examination were invited to participate. They which is about 45% or more above the old insurance comprised 3155 men (85%), the remaining having standard.9 We made a complete search of the files and moved, emigrated, disappeared or died, or their addresses were unknown. Those selected for the follow found 1940 men fulfilling this criterion. up examination were invited by letter to participate in a health examination performed by the Copenhagen CONTROL GROUP Heart Study at the Rigshospital.4 A reply slip was 1 A control group was derived from a random % City to the letter of invitation urging the recipient attached sample of the study population. With the exclusion of to indicate whether or not he wished to attend the were men for whom height and weight measurements examination. was sent 3 months after the not available and the severely obese (n =21) already first invitationAtoreminder those who had either not responded control the group included in the obese group, The were divided into to of failed appear. participants comprised 3601 men. For the present purposes, we seven regional groups according to proximity selected a random 50% sample of this control group. current residence: region 1 comprisedof their the municipalities closest to the hospital and region 7 DRAFT BOARD EXAMINATION comprised municipalities up to 100 km from the place When appearing before the medical board, the young of examination. men are measured and weighed wearing their underclothes only. Since 1958 the examination has HOSPITAL ADMISSION included an intelligence test (B0rge Prien's test, 1953) Information regarding the respondent's admission to and an assessment of educational level. The test is a 45 every Danish somatic hospital during the period minute paper and pencil examination for groups and 1977-82 was obtained from National Health Service its principles and properties have been described Hospital Admission Register.'4 previously.'o It comprises four subtests-letter matrices, verbal analogies, number series, and STATISTICAL METHODS geometric figures-with a total of 78 items. The test The x2 test and the Goodman-Kruskal test (gamma) score was the total number of correct answers. for trends in two and three dimensional contingency Educational level was recorded on a nine point scale, tables were used.'5 The combined influence on primarily indicating years of schooling from 7 through response rate of body mass index, intelligence test 12 years. It also incorporated recognition, at lower score, education, social class, age and residence levels, of some post-school training, and, at higher distance of the invited subjects was evaluated by levels of success in public examinations including logistic regression analysis.'6 academic degrees." Results SOCIAL CLASS

Social class was derived from the subjects' occupations taken from the annual income tax returns. By use of the National Population Register, the occupational title of all subjects was obtained by November 1, 1980. Social class was rated on an eight point scale from 0 (low: unskilled, manual worker, eg, cleaner, porter) to 7 (high: advanced professional positions, eg, university professor, judge) according to a

BODY MASS INDEX

The response rate was 75-4% (1134/1504) in the control group and only 58X3% (964/1654) in the obese group (p < 0.000, x2 test). For both weight groups there was an increasing rate of non-response with increasing body mass index at the time of appearance before the medical board (p